海洋钻井平台中Q690E焊材的国产化

用于大型浮吊等海工产品的Q690E低合金高强钢焊接时易产生冷裂纹,热影响区受加热速度和冷却速度的影响易形成软化带和脆性组织。焊接时应采用焊前预热、焊后保温和较小的热输入。对比分析国产焊材和国外焊材的化学成分、熔敷金属的力学性能和焊后焊缝金属的力学性能,在可靠的焊接工艺规范下,国产焊材是完全能够替代进口焊材的。     

S模式应答信号多通道接收解码技术研究

为了解决单个天线覆盖范围和作用距离之间的矛盾，实现更广的空域覆盖和更高解码率的目标，采用多个通道进行S模式应答信号的接收、解码及融合纠错的方法，进行了两通道接收、解码及融合纠错实验，获得具有代表性的三批目标的解码数据。经过统计分析，结果表明多通道接收解码不仪可以覆盖更广的空域范围以增加数据量，同时在其中两个通道覆盖的交叠区域，采用信息融合纠错的方法可提供较好的相互纠错能力。     

A numerical methodology for optimizing the geometry of composite structural parts with regard to strength

The increased use of composite materials in lightweight structures has generated the need for optimizing the geometry of composite structural parts with regard to strength, weight and cost. Most existing optimization methodologies focus on weight and cost mainly due to the difficulties in predicting strength of composite materials. In this paper, a numerical methodology for optimizing the geometry of composite structural parts with regard to strength by maintaining the initial weight is proposed. The methodology is a combination of the optimization module of the ANSYS FE code and a progressive damage modeling module. Both modules and the interface between them were programmed using the ANSYS programming language, thus enabling the implementation of the methodology in a single step. The parametric design language involves two verifications tests: one of the progressive damage model against experiments and one of the global optimization methodology performed by comparing the strength of the initial and the optimum geometry. There were made two applications of the numerical optimization methodology, both on H-shaped adhesively bonded joints subjected to quasi-static load. In the first application, the H-shaped joining profile was made from non-crimp fabric composite material while in the second from a novel fully interlaced 3D woven composite material. In the optimization of the joint’s geometry, failure in the composite material as well as debonding between the assembled parts was considered. For both cases, the optimization led to a considerable increase in joint’s strength.     

CAST工艺处理市政污水的设计与运行

大连老虎滩污水处理厂采用CAST工艺处理市政污水,在处理水量超出设计值12.5%、进水浓度低于设计值的条件下,通过调整运行参数,使出水水质达到《城镇污水处理厂污染物排放标准》（GB 18918—2002）的一级B标准,综合运行电耗为0.126kW·h/m3,去除污染物单位电耗为2.73kW·h/kgBOD。     

Assessing and optimizing the economic and environmental impacts of cogeneration/district energy systems using an energy equilibrium model

Energy equilibrium models can be valuable aids in energy planning and decision-making. In such models, supply is represented by a cost-minimizing linear submodel and demand by a smooth vector-valued function of prices. In this paper, we use the energy equilibrium model to study conventional systems and cogeneration-based district energy (DE) systems for providing heating, cooling and electrical services, not only to assess the potential economic and environmental benefits of cogeneration-based DE systems, but also to develop optimal configurations while accounting for such factors as economics and environmental impact. The energy equilibrium model is formulated and solved with software called WATEMS, which uses sequential non-linear programming to calculate the intertemporal equilibrium of energy supplies and demands. The methods of analysis and evaluation for the economic and environmental impacts are carefully explored. An illustrative energy equilibrium model of conventional and cogeneration-based DE systems is developed within WATEMS to compare quantitatively the economic and environmental impacts of those systems for various scenarios.     

基于深度学习的水面漂浮物智能检测方法

漂浮物作为河道表观污染的重要源头,加强漂浮物检测是改善水环境生态质量的重要途径,也是积极落实“河长制”政策的技术手段。由于水面漂浮物具有场景复杂度高、形状不规则以及多尺度形态变化等特点,采用传统的图像识别方法快速有效地检测目标具有较大的挑战性。因此,本文提出了一种基于深度学习的实时且稳健的水面漂浮物智能检测方法。首先,基于稀疏分解思想对低质量漂浮物图像进行降噪和增强处理,初步降低复杂水面环境对漂浮物图像质量的影响。其次,将轻量化MobileNetV2网络取代SSD算法中的VGG16网络作为骨干网络,在预测层中将深度可分离卷积取代标准卷积,并采用动态特征金字塔网络提高多尺度漂浮物的检测精度,弥补SSD网络中强制不同层学习相同特征的不足。然后,将统一量化卷积神经网络（Quantized-CNN）框架应用于量化SSD检测器的检测误差,进一步加速卷积层计算和压缩全连接层的参数,降低SSD算法的计算复杂度和内存成本。本文在构建的水面漂浮物数据集上进行的实验结果表明：与现有的图像识别算法相比,改进后的SSD检测算法的平均精度（AP）和F1达到95.86%和94.74%,在硬件GPU下的检测速度达到64.23f/s,检测算法的参数计算量减少到7.5亿,模型内存成本压缩到6.27MB。改进SSD算法实现了水面漂浮物检测的高精度和高效率。     

几何引导的主动式三维感知与交互

随着三维感知设备的发展和大规模三维数据的出现,基于三维重建与理解的视觉感知技术得到了大量关注。与此同时,智能图形逐渐改变了传统图形系统在交互中的被动角色,朝着任务引导的、感知驱动的智能体对真实或虚拟环境的主动交互发展。可以说,计算机图形学正在突破“信息表达”这一传统范畴,逐步拓展迈入“信息感知”领域;图形学的交互技术也由传统的人机交互,逐渐延伸和发展出面向智能任务的主动三维交互。其中,数据驱动三维几何分析与建模的理论和方法,特别是在线重建与分析技术,对三维感知和三维交互形成了重要支撑。本文从图形学和视觉融合的视角,结合研究案例,介绍了主动式三维感知与交互,讨论了“主动式”的特点、优势和挑战,并试图探讨这一方向的开放问题与发展趋势。     

锦屏二级水电站施工排水洞岩爆数值模拟

岩爆是一种极为复杂的动力失稳现象.在深埋隧洞的勘察设计和施工过程中,高地应力及其诱发的频繁岩爆,是影响隧道洞室稳定的重要因素之一.作为四川雅砻江锦屏二级水电站的重要的临时工程,施工排水洞在进入埋深1600m后就发生多次岩爆, 严重制约了工程的进展,为了研究岩爆发生机理与高地应力之间的关系以及更好地预测岩爆在隧洞内发生的位置,运用渐进破坏过程数值分析软件RFPA(realistic failure process analysis)对施工排水洞进行初步的应力分析,得出与现场实测相吻合的初始地应力拟合曲线.同时,隧洞开挖引起应力重分布而产生的围岩破坏的数值模拟也很好的印证了现场发生的岩爆情况,表明基于渐进破坏过程数值分析的RFPA可以真实地模拟高地应力下岩爆机理.     

纳米粒子的结构对EVOH纳米复合物性能的影响

为研究纳米填充物对EVOH基体性能的影响,采取熔融共混的方法,用EVOH-56作为基体,按照不同比例填充了化学结构或聚集态结构不同的纳米粉体,从而制备出不同的EVOH纳米复合体系。根据TPT方程,计算出与复合体系中各组分间相互作用相关的参数B值,EVOH-56/纳米TiO_2复合材料的B值(16.493)大于EVOH-56/纳米ZnO复合材料的B值(3.806);随着纳米粉体含量的增加,两种复合物体系的熔体温度升高约6%;而在相同纳米粉含量条件下,EVOH-56/纳米ZnO体系的加工扭矩比EVOH 56/纳米TiO_2体系低5%～10%;在力学性能方面,EVOH-56/nano-ZnO体系的屈服强度值和弹性模量值均优于EVOH 56/nano-TiO_2体系。     

应对区域供电线路故障的多功能复合储能优化配置方法

区域供电线路故障会使部分线路出现短时拥塞,同时区域内重要负荷供电不足,导致大量切机和切负荷。针对这一问题,提出一种应对区域供电线路故障的多功能复合储能优化配置方法。该配置方法考虑了各发电单元在线路故障发生后的响应时间,优化配置功率型储能和切除部分非关键性负荷以缓解线路短时拥塞。考虑重要负荷在故障检修期间的供电需求指标,在已有功率型储能基础上,协调优化配置能量型储能以保障区内重要负荷供电。最后,以复合储能整体投资费用最小为优化目标,结合实际电力系统算例进行了仿真,对影响配置结果的关键参数进行了灵敏度计算,深入分析了复合储能优化配置的相关原理,验证了所提模型和方法的正确性。